Parental overprotection is thought to be a risk factor for childhood anxiety disorders. The Parental Overprotection Measure (POM) is a self-assessment scale that has been used in many studies, however, no article has comprehensively explored its psychometric properties. The aim of this study was to evaluate the POM to secure reliable, replicable, and comparable results. Using an item-response theoretical framework, we aimed to address five basic criteria of psychometric quality: one-dimensionality, ordered response categories, invariance, targeting, and reliability. Furthermore, we aimed to identify and exclude any items that did not contribute significantly, to create a concise and practical scale. A total of 1092 parents of children aged 4–12 years were recruited through an online advertisement and completed an anonymous online survey during September and October 2020. Out of the 19 items of the original scale, eight were excluded due to unsatisfactory psychometric properties and/or because of not being judged by experts as harmonious with the concept of parental overprotection. The retained items constituted the POM-11, a brief scale with sound psychometric properties that has the potential for use in both research and clinical settings.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
According to theoretical models, parental overinvolvement can lead to increased levels of anxiety in children (Chorpita & Barlow, 1998; McLeod et al., 2007; Rubin & Mills, 1991). In these models, it is postulated that a child who is repeatedly controlled or protected where independence is developmentally appropriate, can be hindered from developing a sense of mastery in challenging situations, and therefore becomes more anxious (Chorpita & Barlow, 1998; McLeod et al., 2007). In contrast to parental overinvolvement stands parental autonomy granting. When the parent encourages the child to be independent in novel situations, it is believed that the child gets the opportunity to develop a sense of control of the situation, which can reduce the risk for anxiety (Becker et al., 2010; Wood et al., 2003).
Several systematic reviews indicate that child anxiety symptoms are associated with parental overinvolvement and lack of autonomy granting (McLeod et al., 2007; Wood et al., 2003; Yap & Jorm, 2015). However, the possibilities to draw conclusions from these compilations are limited because parental overinvolvement encompasses a variety of parenting behaviors. To reach a better understanding of the link between parenting and childhood anxiety, there is a need to breakdown the higher order construct of parental overinvolvement to more specific, lower order parenting dimensions. This is highlighted by the results from a meta-analysis conducted by (Möller et al., 2016) which included mainly cross-sectional studies with young children (ages 0–5 years). Möller et al. (2016) used the term parental overinvolvement as a broader concept, including the subdimensions overprotection and overcontrol. Parental overprotection was defined as being overly worried about the safety and wellbeing of the child, behaving in an overly cautious manner towards the child, and limiting the child’s exposure to unfamiliar situations and persons. Parental overcontrol, on the other hand, was defined as the parent excessively helping or interfering with the child’s behavior, without taking the child’s needs or interests into consideration. The result showed that parental overcontrol was not related to child anxiety, however, parental overprotection showed a small significant association with childhood anxiety. The authors conclude that the examined parenting behaviors played only a small role in explaining anxiety in young children and raise the possibility that the effect of parenting on child anxiety may accumulate over time.
Parental overprotection can manifest as warning the child about minor dangers, fostering the child to stay dependent on the parent and responding in a needlessly solicitous way when the child is anxious (e.g., letting the child avoid anxiety triggering situations). Most research on the relationship between parental overprotection and child anxiety is correlational in nature, but a few longitudinal studies have been conducted on this topic. For example, in a study conducted by (Edwards et al. 2010), mothers and fathers of preschool-aged children, completed questionnaires with 12 months apart. It was found that both maternal and paternal scores on The Parental Overprotection Measure (POM) predicted child anxiety one year later. Further, a bidirectional relationship was found between maternal (but not paternal) overprotection and child anxiety, where maternal overprotection both predicted and was predicted by child anxiety. Of note, in a succeeding longitudinal study with a similar design, parents of children aged 3 to 6 completed questionnaires twice with a one-year interval (Vreeke et al., 2013). The results showed that parent self-rated overprotection did not predict level of child anxiety twelve months later.
There are several measurements available that assess different aspects of the broader construct parental overinvolvement (for a review, see Lohman & Bayer, 2020). One measurement that specifically evaluates overprotection is the POM, a self-assessment scale administered to parents. Edwards et al. (2008) developed the scale and validated it in a sample of 636 mothers and 251 fathers of 3–5-year-old children. The results showed that the POM had high levels of internal consistency and high test-retest reliability. POM-scores were further found to be significantly correlated with observer ratings of maternal overprotective behavior during a laboratory task (Edwards et al., 2008). Although the POM has been used in many studies (Edwards et al. 2010; M. Howard et al., 2017; Pizzo et al., 2022; Vreeke et al., 2013), its psychometric properties has not been comprehensively explored. Despite the limited availability of psychometric data for the POM, the measure possesses significant advantages. It distinctly focuses on overprotection, as opposed to overcontrol. Further, the items emphasize parental behaviors or reactions in different situations, rather than attitudes. The scale includes statements like “I do everything possible to protect my child from potential injury”,” I try to protect my child from making mistakes” and “I accompany my child on all outgoings”. Furthermore, it has been used with parents of children in both early (Edwards et al., 2008) and middle childhood (Clarke et al., 2013), suggesting it may be a suitable measure for longitudinal studies.
Objectives
The primary goal of this study was to evaluate the psychometric properties of the POM to secure reliable, replicable, and comparable research results. Using an item-response theoretical framework, we aimed to address five basic criteria of psychometric quality assessment as proposed by Johansson et al. (2023): one-dimensionality, ordered response categories, invariance, targeting, and reliability. Furthermore, we aimed to identify and exclude any items that did not contribute significantly, or that distorted the scale. This step was taken with the intention of creating a concise and practical scale, making it well-suited for both research and clinical settings.
Method
Participants
A Facebook advertisement was used to recruit study participants. The only requirement to be a part of the study was to have a child between the ages of 4 and 12 years. No specific exclusion criteria were used. A total of 1092 individuals completed the survey. To prevent multiple submissions, the survey was configured to allow only one response per device. In addition, we included two pre-screening items to check respondents’ attention and eligibility (Curran, 2016). The first, “Do you have a child aged 4–12 years?” (“Yes”/“No”), ensured the correct target population, while the second confirmed respondents’ age and informed consent. Respondents who failed either check were automatically excluded. The demographic characteristics of the sample are presented in Table 1. The study included slightly more mothers than fathers, with mothers comprising 55% of the participants. Among the children involved in the study, the sex distribution was even (50% boys and 48% girls). The mean age of the parents was 41.0 years (SD = 5.6) and the mean age of the children was 8.0 years (SD = 2.5). The survey also included questions about previous mental health care. Among the parents, 44% responded that they had at some point sought professional help for mental health issues. Regarding the participants’ children, 16% of them had received professional help for a mental health issue or developmental disability at some point.
Table 1
Descriptive statistics of participants and their children
Characteristic
n
% of sample
Parent gender
Female
600
55%
Male
484
44%
Other
8
1%
Parent education level
Upper secondary school (≤12 years)
203
19%
Post-secondary education (>12 years)
889
81%
Parent country of birth
Sweden
1009
92%
Abroad
83
8%
Child gender
Female
529
48%
Male
548
50%
Other
15
1%
Parent current or previous mental health care
No
616
56%
Yes
476
44%
Child current or previous care related to mental health or developmental disabilities
No
915
84%
Yes
177
16%
Procedure
Parents who clicked on the advertisement were directed to Karolinska Institutet’s webpage, where information about the study was provided. In the next step, interested parents were directed to a completely anonymous online survey (i.e., we did not collect any personal data). Before proceeding with the survey, parents were required to provide consent. This was done by acknowledging their agreement through a checkbox. The survey was live for a period of three weeks, during September and October 2020.
Measures
The online survey comprised general demographic questions, the Parental Overprotection Measure (POM) and PROMIS anxiety short forms. Additionally, an item pool containing 51 statements covering different aspects of parental modeling of anxious and non-anxious behaviors were included in the survey. These items were generated as part of the process to develop a new questionnaire measuring parental anxious modeling, a process described in a previous publication (Elfström & Ahlen, 2022).
Parental overprotection measure (POM)
The POM (Edwards et al., 2008) is a questionnaire designed to assess parents’ tendencies to engage in overprotective behaviors. It comprises 19 items aimed at evaluating parenting behaviors that limit a child’s exposure to perceived dangers or risks. Parents are requested to assess the degree to which each statement reflects their usual response on a 5-point scale, ranging from 0 (not at all) to 4 (very much). In a sample of parents of children aged 3–5, Edwards et al. (2008) found high twelve-months test-re-test reliability and high internal consistency for both informants. In a sub-sample of mothers, it was further found that POM-scores significantly correlated with observed maternal behavior on a challenging physical task. In a subsequent study with a sample of mothers of children aged 7–12 years, the POM showed good internal consistency of scale scores, and a small to moderate correlation between total scores and external ratings of maternal overprotection provided by observers (Clarke et al., 2013).
PROMIS anxiety short forms
The Patient-Reported Outcomes Measurement Information System (PROMIS) is an initiative led by the National Institutes of Health. PROMIS encompasses various mental and physical health aspects and consists of validated item banks and short forms that were developed using Item Response Theory (IRT) (Pilkonis et al., 2011). In this study, we employed the PROMIS anxiety short form (eight items) for adults, which we will refer to as PROMIS-parent, to assess parental anxiety (Pilkonis et al., 2011). In our sample, the Cronbach’s alpha coefficient for PROMIS-parent was 0.93. To measure child anxiety, we utilized the PROMIS parent proxy report scale (eight items), referred to as PROMIS-child (Varni et al., 2012). The Cronbach’s alpha coefficient for PROMIS-child was 0.86. For both scales, respondents rated the items on a frequency scale ranging from (1) never to (5) always.
Being on guard
Being on guard is a self-assessment scale for parents, with nine items answered on a frequency scale from (1) never to (5) almost always. The scale describes behaviors where the parent is displaying anxiety in front of the child by acting to avoid harm (i.e., “I am on my guard and attentive to danger when me and my child are out together” and “I tell my child to be careful.”). The behaviors described in this scale could be seen as indicative of both overprotection and the modeling of anxiety. According to a one-factor EFA of Being on guard, all nine items loaded strongly on the factor and according to Rasch partial credit model, all items fitted the model well (Elfström & Ahlen, 2022). The Cronbach’s alpha in this sample was 0.78.
Data Analysis
All statistical analyses was performed in the R software (R Core Team, 2022) using the “psych”, “Hmisc” and the “RISEkbmRasch” packages.
Step 1 – Rasch Analysis
We used a polytomous Rasch model (the Partial Credit Model) to analyze the POM. Rasch is an adequate model when exploring scales intended to be interpreted as the summation of scores (Hagquist, 2007). A central aspect of a scale is that the items measure one single construct (McNeish & Wolf, 2020). We explored the standardized Rasch residuals in a principal component analysis (PCA). A one-dimensional measure should not produce residuals that are strongly correlated, otherwise there is evidence of further underlying constructs. An eigenvalue exceeding 2.0 in the first component of a PCA has been proposed as an indicator that the assumption of one-dimensionality is not met (Boone & Staver, 2020).
Using the Rasch model, we explored each item’s thresholds. In a one-dimensional measure, the responses to the ranked categories (for each item) should increase as the underlying trait increases. If item thresholds are disordered (i.e., the estimated thresholds are not in the same order as the item categories) this may indicate difficulties with the item (Hagquist, 2007). In addition, we inspected item targeting, which involves assessing whether the thresholds of response categories align with the positions of individuals in the population along the underlying trait (Boone & Noltemeyer, 2017).
Furthermore, we explored each item’s fit to the Rasch model. We calculated the mean square statistics (MSQ) which is the average of all participants’ squared standardized differences between observed responses and predicted responses according to the Rasch model. An MSQ above 1.3 indicates that the item does not fit the model very well while an MSQ below 0.7 indicate that the item may not contribute much, i.e., being redundant (Smith et al., 2008).
Finally, we examined differential item functioning (DIF) to investigate whether item difficulty was consistent across different groups (Muraki, 1999). We studied DIF between mother- and father ratings, ratings of girls and boys, and ratings of pre-school children (4–6 years old) and school children (7–12 years old). Differences in item difficulty between groups may suggest issues with the item, implying that despite two individual share the same level of the latent trait, they achieve different scores due to their group affiliation (Wu et al., 2016). DIF of 0.3 has been suggested to indicate small DIF, 0.5 moderate DIF, and 0.7 large DIF (Jin et al., 2012).
Step 2 – Factor analysis
In addition to the Rasch analyses, we investigated the POM in an exploratory factor analysis (EFA) using maximum likelihood factoring extraction method (Costello & Osborne, 2005). Only a one-factor model was tested. The EFA utilized the polychoric correlation matrix to ensure accurate factor solutions when working with Likert scales (Holgado–Tello et al., 2008). Loadings above 0.40 were considered satisfactory (Howard, 2016).
Step 3 – Qualitative assessment and item reduction
Each item was rated independently by three experts regarding whether the item was (a) harmonious with the concept of overprotection, (b) may imply overprotection, (c) not relevant or lacking clear implications of overprotection. Experts were all clinical psychologists and researchers within the field of child psychology and parenting interventions. All experts were born in Sweden. Based on the input from the experts, and our own review of each item, we concluded on what items that may be problematic to include in a final set of items. An integrated assessment of the results from the EFA, Rasch analysis (i.e., ordered thresholds and targeting, item fit, and DIF) together with the results from the qualitative assessment was considered when determining what items to retain in the final item set. For the final scale we explored items in a PCA of residuals, residual correlations, item thresholds and targeting, item fit, DIF, and a one-model EFA.
Step 4 – Reliability
For the final scale, we examined reliability using the Test Information Function (TIF) curve, which illustrates how reliability changes across different levels of the latent trait (Samajima, 1994). A TIF value exceeding 3.33 corresponds to a reliability of 0.7 (on a scale from 0 to 1), often considered the threshold for acceptable reliability across various assessment metrics. For comparison to other studies, we additionally calculated Cronbach’s alpha for the final scale, and the original 19-item POM.
Step 5 – Association to other measures
In a previous article based on the same sample, we presented correlations between the original POM and the being on guard subscale, parental anxiety and child anxiety (Elfström & Ahlen, 2022). Guidelines suggest that a correlation ≥0.1 is a small correlation, ≥0.3 is a moderate correlation, and ≥0.5 a strong correlation (Cohen, 1988). The original POM showed a strong correlation to the being on guard subscale, which is adequate as the being on guard subscale measures overprotective behaviors performed in front of the child. Further there was a moderate association with parental anxiety, which is coherent as anxious parents tend to be more overprotective compared to non-anxious parents (Jones et al., 2021). Finally, there was a small association with child anxiety, which is also consistent with previous research (Möller et al., 2016).
To assess whether changes to the POM scale resulted in any loss in validity, we compared the final scale with the original 19-item POM in terms of their associations with the ‘being on guard’ scale, parental anxiety, and child anxiety. To statistically explore differences between dependent correlations we used the William’s test (Steiger, 1980).
Results
Initial Data Quality
We conducted a post-hoc long-string analysis on the 19-item POM scale to identify careless responding patterns. A long-string analysis examines the longest sequence of identical responses for each participant (Curran, 2016). Of the 1092 participants, 89 (8%) had a long-string greater than five identical responses. Most (79) of these were strings of “0” concentrated around items 13–15, items with high difficulty (Supplementary Fig. S1). As these patterns likely reflected genuine difficulty rather than inattentiveness, no participants were excluded in the analysis.
Rasch and Factor Analysis
A PCA of Rasch residuals on the original POM showed an eigenvalue of 2.21, indicating some problems with the one-dimensionality assumption. Eleven items showed disordered thresholds. Five were concluded as marginally disordered (i.e., thresholds overlapped, or were marginally reversed), and six items showed clearly reversed thresholds. The six items with clearly disordered categories are presented in Table 2. Overall, targeting of items was adequate with items thresholds across all levels of the underlying trait (Supplementary Fig. S1 presents ordering of thresholds for all items).
Table 2
Assessment of item suitability using psychometric tests and qualitative assessment
Item summary
Rasch analysis
EFA
Qualitative assessment
Disordered thresholds
Misfitted item
DIF
Weak factor loadings
Problem Expert 1
Problem Expert 2
Problem Expert 3
1. Comfort immediately
X
X
X
X
X
2. Keep in close distance
X
X
3. Protect from criticism
4. Attention when clingy
X
X
5. Not let child out without me
X
X
6. To doctor if unwell
X
X
X
7. Keep a close watch
8. Overprotective
9. Foresee and avoid risks
X
10. Protect from mistakes
11. Not climb trees
X
X
12. Shelter from troubles
13. Panic if out of sight
14. Reluctant about risky sports
15. Not leave child to others
X
16. Accompany on outgoings
X
17. Shield from conflicts
18. All to prevent injury
19. Protect from child’s fears
Items marked with an ‘X’ indicate issues identified by specific psychometric tests or assessment
EFA exploratory factor analysis
According to item fit, four items showed MSQ values above 1.3 indicating that they did not fit the model well (Table 2). No items were categorized as redundant (Supplementary Table S1 presents MSQ values for all items). The DIF analyses showed potential problems with two items regarding differences in difficulty between pre-school children and school children (see Table 2). Supplementary Table S2 illustrates Differential Item Functioning (DIF) for each item across the three comparisons.
In the EFA, using a one factor model, three of 19 items showed unsatisfactory loadings below 0.4 (Table 2). Supplementary Table S3 presents factor loadings for all items.
Qualitative Assessment
Nine items were unanimously rated by the experts as harmonious with the concept of overprotection. Regarding six items, the experts disagreed on whether they were in harmony with or only may suggest overprotection. Finally, the expert ratings differed in terms of whether four of the items were harmonious with, may imply or were not relevant/lack clear implications of overprotection (see Table 2 and details in Supplementary Table 4).
After an integrated assessment of the ordered thresholds and targeting, item fit, DIF, and EFA together with the results from the qualitative assessment, eleven items were decided to be included in the final set of items (See items displayed as bold in Table 2). All items displaying psychometrically unsound properties were excluded. Although one item “I try to anticipate and avoid situations where my child might do something risky” was considered by one expert to be not relevant of overprotection, we decided to retain this item, based on its sound psychometrical properties.
When exploring the 11 remaining items in a Rasch model, no item showed misfit (Supplementary Table S5), no items showed problematic DIF (Supplementary Table S6), and all items had satisfactory loadings in a one factor EFA (Supplementary Table S7). A PCA of residuals revealed an eigenvalue of 1.98, slightly below the threshold that indicates potential issues with one-dimensionality due to residual correlations. Examining the residual correlation matrix, we found that four pairs of items displayed residual correlations exceeding the average correlation by 0.2, a level indicating possible concerns (Christensen et al., 2016). Nevertheless, we opted not to exclude additional items solely based on this criterion. Supplementary Table S8 presents the residual correlation matrix.
Reliability
The items in the final version of the POM (hereafter called POM-11) showed very good targeting in relation to the study population. As presented in Fig. 1, item thresholds were distributed across the population distribution, indicating that the scale can differentiate between individuals across different levels of the underlying construct. According to the TIF-curve, the POM-11 showed acceptable reliability (>3.3) in 98% of the sample. Cronbach’s alpha for the POM-11 was 0.85. The original POM showed acceptable reliability (>3.3) in 100% of the sample. Cronbach’s alpha for the original POM was 0.86.
Fig. 1
Person-Item Map of the POM-11 showing item thresholds (and averages) in relation to the latent dimension and population distribution
×
Associations to Other Measures
Table 3 shows associations between the POM-11, the original POM, the being on guard subscale, parental anxiety, and child anxiety. As expected, a strong correlation, r = 0.57, was found between the POM-11 and the being on guard subscale. The correlation was statistically stronger than the correlation between the original POM and the being on guard subscale (p < 0.001). Furthermore, a moderate correlation, r = 0.34, was found between the POM-11 and parental anxiety. The correlation was not statistically different than the correlation between the original POM and parental anxiety (p = 0.23). Finally, a small correlation was found between the POM-11 and child anxiety (r = 0.20). The correlation was statistically stronger than the correlation between the original POM and child anxiety (p = 0.006).
Table 3
Pearson correlations between included variables
POM-11
Being on guard
Parental anxiety
Child anxiety
POM
0.95
0.55
0.33
0.17
POM-11
0.57
0.34
0.20
Being on guard
0.25
0.17
Parental anxiety
0.36
POM parental overprotective measure, POM-11 11 item version of the parental overprotective measure
Discussion
This study aimed to assess the psychometric properties of the POM. Out of the 19 items of the original scale, eight were excluded due to unsatisfactory psychometric properties and/or because of not being judged by experts as harmonious with the concept of parental overprotection. The retained items constituted the POM-11, a shorter scale with adequate psychometric properties that has the potential for use in both research and clinical settings.
When items were excluded from the original POM, information relevant to the construct of parental overprotection might have been lost. However, these concerns can be alleviated by the fact that POM-11 covers a wide range of overprotective parenting behaviors. Importantly, the short scale includes both parents’ tendencies to shield their child from physical harm (e.g., I am hesitant to let my child participate in certain sports out of fear that he/she might get injured) and emotional distress (e.g., I protect my child from his/her fears). While lengthy measurements have historically been seen as more reliable and valid, Smith et al. (2012) have suggested that brief tests can provide valid assessments of specific constructs. Smith et al. (2012) reasoned that to achieve content validity, it is not enough to solely identify what content is part of the target construct, but that it is just as fundamental to identify what content is not related to the construct of interest, and hence to remove any non-exemplary content. In the present study, in alignment with this perspective on content validity, statistical analyses and experts’ ratings of the items’ relevance to parental overprotection were considered when determining which items to retain. Overall, we found that there were no evidence of loss of validity or reliability for the POM-11 compared to the original scale. We found that the POM-11 had a small but significant association with child anxiety symptoms. This is in line with much of the previous research, where parental overprotection have been found to be associated with child anxiety symptoms in both cross-sectional (McLeod et al., 2007; Möller et al., 2016) and longitudinal studies (e.g., Edwards et al. 2010). However, this result differs from the findings of Clarke et al. (2013), where no association between total POM scores and child anxiety symptoms was found. Clarke et al. reasoned that this might be explained by the higher age of the children in their study compared to previous research where POM has been used. Although this explanation cannot be ruled out, there are also other possible reasons why no association was observed. The study had a small sample size (n = 90), which does not allow detection of associations in the smaller range. Specific parenting behaviors seem to only explain a small percentage of variance in child anxiety symptoms (McLeod et al., 2007). Hence, the examination of the relationship between parenting and child anxiety is more comprehensively explored in larger sample sizes. Further, it is possible that shortcomings in the measurement method (i.e., the original POM) obscured a true relationship between child anxiety and parental overprotection. In the current study, a considerable number of items in the original POM were found to have inadequate psychometric qualities, which raises some questions about the possibility to draw accurate conclusions from prior studies that have used the full scale.
The POM-11 scores were found to be moderately associated with parent anxiety symptoms. This is in line with the results from a recent review investigating the association between maternal anxiety and overprotection. Jones et al. (2021) identified 13 studies investigating the subject and concluded that the studies with the highest quality gave support for an association of small to medium size. It is important to highlight that parental anxiety can only partly explain the variance in overprotective behaviors in parents. Other factors, especially child temperament, has been suggested to lead to increased levels of parental overprotection (Lewis-Morrarty et al., 2012), and researchers has highlighted a possible bi-directional relationship between parental overprotection and child anxiety (Edwards et al., 2010). Hence, to address parental overprotection is not only relevant when working with anxiety disordered parents, it can also be a relevant aspect to consider in any families with children with anxiety disorder where the parents do not have clinical levels of anxiety. Of note, most of the previous research has focused specifically on maternal overprotection, while neglecting the role of paternal overprotection. Therefore, a strength of our study is that it included both mothers (55%) and fathers (45%), resulting in that POM-11 is adequate to use as a measure of parental overprotection irrespective of parent gender. Additionally, POM-11 was found to be strongly correlated with the subscale “Being on guard”. This outcome was anticipated since most items within this subscale describe parenting behaviors that aim to shield the child from danger.
When interpreting the results, it should be considered that we conducted the current study during the COVID-19 pandemic, which may have heightened the perceived need to protect one’s child. The study was conducted in Sweden, where society did not shut down to the same extent as in many other countries, with schools remaining open and fewer restrictions in place. Nevertheless, the pandemic could have amplified parents’ overprotective tendencies, and hence influenced participants’ responses. Most of the questions focused on specific situations which are unlikely to have been directly affected by the pandemic (e.g., protecting a child from criticism, reluctant about risky sports and protects from conflicts), while the answer to other more general questions (e.g., foresees and avoids risks) are more likely to have been influenced by the pandemic.
Limitations
Smith et al. (2000) have underscored that when developing a short version of an existing scale, it is necessary to demonstrate reliability and validity on the short form separately from what has been demonstrated on the original form. In the current study, the original POM-scale was administered to the participants. When POM-11 is administered separately, the adjacent items will differ, and we cannot know if this will lead to any differences in the response patterns. To further ensure the validity and reliability of the POM-11, it would therefore be of value to conduct a subsequent psychometric evaluation where the short scale is administered as it is.
As pointed out by Podsakoff et al. (2012), the associations between many constructs will differ significantly depending on if data is collected from a single source or from multiple sources. Hence, a limitation of the current study is that it relies on parent report only, meaning that there is a risk for common method bias (MacKenzie & Podsakoff, 2012). If instead both parent, child, and observer report had been used in the current study, the result might had been different.
All the experts involved in the qualitative assessment of the item relevance were born in Sweden, which limit the cultural applicability of their assessments. Parenting behaviors and norms vary across cultures (Lin et al., 2023), and the lack of expert diversity may affect the generalizability of the results of the current study. Future research should involve experts from more diverse backgrounds to ensure broader applicability across different populations.
Another limitation is that the participants in this study exhibited a higher level of education compared to the broader Swedish population, potentially limiting the extent to which the results can be generalized to other sub-groups. Additionally, since the sample was drawn from Facebook, self-referral may have led to the recruitment of a specific group of participants that are interested in research and parenting, which may not fully represent the general population. Furthermore, 92% of the participants were born in Sweden, meaning that the findings may not be representative of families with diverse cultural backgrounds, limiting the generalizability of the results. The POM-11 should be examined in other populations to assess its applicability across different demographic groups. Additionally, the fact that we cannot know for certain how the timing of the study (i.e., during the pandemic) may have impacted participants’ responses might also limit the generalizability of the results.
Conclusions
In summary, we developed POM-11, a brief measure with sound psychometric properties. POM-11 has the potential for being used in research as well as clinical settings. It can be used both in basic research to understand the relationship between anxiety and overprotection, and in intervention studies, either as an outcome measure or to assess mediation. Clinically, POM-11 can be used to identify overprotective behaviors in parents, allowing clinicians to address these tendencies when working with families where parents and children suffer from anxiety. However, before using POM-11, we recommend that researchers and clinicians consider the population in which the psychometric evaluation was conducted.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Met BSL Psychologie Totaal blijf je als professional steeds op de hoogte van de nieuwste ontwikkelingen binnen jouw vak. Met het online abonnement heb je toegang tot een groot aantal boeken, protocollen, vaktijdschriften en e-learnings op het gebied van psychologie en psychiatrie. Zo kun je op je gemak en wanneer het jou het beste uitkomt verdiepen in jouw vakgebied.