Constructing Measures: An Item Response Modeling Approach

---

A lower value possibly indicated a need for additional items, deleting low corrected item-total correlation items, or the presence of multiple latent trait constructs. The criterion for acceptable corrected item-total correlation was defined as a value greater than 0. An average inter-item correlation between 0.

Exploratory factor analysis was used to assess whether the data satisfied the unidimensionality assumption of item response modeling, and was conducted for each subscale using SPSS IBM Corp, Item difficulty in IRM refers to the level of a latent trait necessary to have a probability of 0. Polytomous item response models are used when items include more than two ordered responses by categories e. For three response options e. A Wright map aligns person performances and item performances on the same interval scale, in units referred to as log odds. A Wright map determines whether the items in a questionnaire cover the range of person abilities in the sample.

Persons are mapped on the left of the map, with lower vegetable parenting practices values located at the bottom of the map. In the present study, the notation x represented the vegetable parenting practices trait estimates, and each x denoted 2. Vegetable parenting practices items and thresholds were located at the right of the map.

Vegetable parenting practices items and thresholds at the top of the scale were more difficult to agree with, becoming easier to endorse further down the scale.

Constructing Measures : An Item Response Modeling Approach

Rating scale model is a simplified version of the partial credit model that assumes the response threshold parameters are identical across items. The infit mean square information-weighted and the outfit mean square outlier-sensitive fit statistics were used to assess item fit. An infit or outfit mean square value of one indicates that the observed variance is similar to the expected variance, while a value close to zero or much greater than one is evidence against the fit of an item. Infit or outfit mean square values greater than 1. For thresholds, outfit mean square values greater than 2.

Differential item functioning is tested by a significant interaction term e. Differential item functioning can be attributed to the presence of nuisance dimensions intruding on the underlying trait that researchers intend to measure Ackerman, While we appreciate the sample size limitation for a three group DIF in these data, each age group is distinct from a developmental psychology perspective Gardner, , and collapsing the age groups would lead to severely unbalanced groups.

A significant chi-square for the item-by-group interaction term, with the ratio of the item-by-group parameter estimates to the corresponding standard error exceeding 1. The differences of the item-by-group interaction parameter estimates were calculated to determine the magnitude of differential item functioning. The parameters were constrained to be zero, so the item-by-group interaction parameter estimates were summed to zero across subgroups. Therefore, the magnitude of differential item functioning difference was twice the estimate of the focal group if only two subgroups were analyzed.

Statistically significant differential item functioning was classified into one of the three categories: The online questionnaire was completed by parents, whose demographic characteristics are shown in Table 1. For the ineffective responsiveness subscale, the second factor had an eigenvalue slightly greater than 1 1.

Therefore, a single common factor was accepted for all the subscales. The internal consistency of each subscale was generally low, but the corrected item-total correlations were acceptable Table 2. Thus, all items were accepted to continue with the analyses. A chi-square test revealed a significant difference effective vegetable parenting practices: Thus, the partial credit model was more appropriate for both effective vegetable parenting practices and ineffective vegetable parenting practices, and was employed in ensuing analyses.

Vegetable parenting practices scale. Item response modeling analyses

A summary of item difficulties and misfit indicators is shown in Table 2. Only one item number 3 did not meet the recommended criterion value of 1. The Wright maps are presented in Fig. Parents with the lowest vegetable parenting practices at the bottom of the scale had difficulty with even the easiest vegetable parenting practices items; parents with greater vegetable parenting practices at the top of the scale had no difficulty performing any of the vegetable parenting practices.

Items became easier to endorse and persons became less able further down the scale. Wright map of effective vegetable parenting practices scale. The notation x represented the vegetable parenting practices trait estimates, and vegetable parenting practices items denoted by numbers and thresholds denoted by number after decimal, for example, Wright map of ineffective parenting practices scale.

The distributions did not fully overlap between item locations and person measures. Items were clustered toward the bottom for effective vegetable parenting practices and toward the upper regions for ineffective vegetable parenting practices. In each subscale category, thus, most parents found it relatively easy to endorse the effective vegetable parenting practices items Fig. Results of the differential item functioning analyses are presented in Table 3.

Parents with girls found it easier to tell their child that vegetables taste good, and ask their child to help with vegetable preparation. However, white parents were less likely to tell their child that their favorite cartoon characters eat vegetables, to give their child something to eat or drink if they were bored, or to admonish their child for not eating their vegetables.

Parents from high income families tended to encourage their child to try a couple of bites of a vegetable, and to reward their child with sweets if they eat their vegetables.

2 editions of this work

Parents with 5-year-olds, more than parents with younger children, tended to tell their child eating vegetables will make them strong and healthy and to admonish their child for not eating their vegetables. Compared with parents with 3-year-olds, parents with an older child showed a higher degree of agreement in encouraging their child to try a couple of bites of a vegetable and in keeping many snack foods in their houses.

Parents with a younger child tended to tell their child that vegetables taste good to a greater extent than did parents with 5-year-olds. Parents with 3-year-olds were more likely to offer vegetables without forcing them to eat them than were parents with 4- or 5-year-olds. Parents with 3-year-olds were more likely to agree with serving meals for the family to eat together and with ineffective vegetable parenting practices items concerning giving their child something to eat or drink if they are upset, and letting their child watch TV at meals.

Compared with parents of the other two age groups, parents with 4-year-olds showed a lower degree of agreement on giving their child something to eat or drink if they are upset. The present study investigated both item difficulty based on multidimensional polytomous item response modeling using the partial credit model which identified misfit items, and examined items exhibiting differential item functioning on the basis of demographic subgroups for a vegetable parenting practices instrument.

One misfit item was identified in the effective vegetable parenting practices scale, and neither effective vegetable parenting practices nor ineffective vegetable parenting practices items covered the full range of the latent trait distribution. Output from the multidimensional partial credit model differential item functioning analyses indicated that some items functioned differently across certain subgroups, even when participants were at the same latent trait level.

Small differential item functioning effects could be ignored, leaving 7 effective vegetable parenting practices and 8 ineffective vegetable parenting practices items with moderate or large differential item functioning. Differential item functioning could be due to differences in the understanding or meaning of the item across groups which could require rewording to minimize these differences , or in the actual use of these vegetable parenting practices, e.

It will be important to modify and re-examine items showing medium and large differential item functioning Angoff, ; Shepard, Revising items should produce non-differential item functioning or considerably lower-differential item functioning items Allalouf, Formative research with samples representative of the target populations may be beneficial to simplify and clarify items. To the extent that differential item functioning reflects true differences in practices, it will be difficult to compare vegetable parenting practices across groups varying in these demographic characteristics using the same scale.

Future research using parent—child interaction observational methods will be needed to clarify the nature of the differential item functioning. Several limitations were identified. No other implementations were identified, and test—retest reliability and inter-rater reliability are not available. The limited overlap between items or item thresholds and individuals on the Wright map indicated that the instruments were not optimally targeted.

Effective vegetable parenting practices items did not cover the more difficult to endorse end of each of the three latent subscale distributions, whereas ineffective vegetable parenting practices items did not cover the easier to endorse end of the distributions. The limited internal consistency reliability and the asymmetric nature of the item and person distributions indicate a need to further develop items. Due to the model complexity of multidimensional polytomous item response modeling, no clear standards for minimum sample size are available. However, diverse minimum sample sizes have been recommended to conduct item response modeling e.

We acknowledged that the relatively small sample size in the present study was a huge limitation. Further research should seek to recruit larger samples and retest these findings. Future research could detect whether the probability of endorsing an item is greater for one group than for the other group over all latent trait levels uniform differential item functioning, e. Additionally, non-uniform differential item functioning -detection procedures e. Psychometric analyses are informative, especially early in instrument development. Limitations of an existing vegetable parenting practices scales were identified.

Parts of the extreme ends of the effective vegetable parenting practices and ineffective vegetable parenting practices construct distributions were not adequately covered by the items. Missing were items that were more difficult and easier to endorse for effective vegetable parenting practices and ineffective vegetable parenting practices, respectively. Additional psychometric work needs to be done; thus, the scales should be used in diverse groups with due caution. This manuscript does not represent the views of the USDA. National Center for Biotechnology Information , U. Author manuscript; available in PMC Aug 1.

The publisher's final edited version of this article is available at Appetite. See other articles in PMC that cite the published article. Abstract Objective To evaluate the psychometric properties of a vegetable parenting practices scale using multidimensional polytomous item response modeling which enables assessing item fit to latent variables and the distributional characteristics of the items in comparison to the respondents.

Method Parents of 3—5 year old children completed a self-reported vegetable parenting practices scale online. Vegetable, Parenting practices, Multidimensional item response modeling, Differential item functioning. Introduction High dietary intake of fruit and vegetables has been associated with reduced risk of cardiovascular disease, stroke, diabetes, and some cancers Boeing et al. Methods General design The study design and methods have been reported in detail elsewhere Baranowski et al.

Refine your editions:

Instrument The survey scale contained 28 items, with 14 items each for effective vegetable parenting practices and ineffective vegetable parenting practices. Table 2 Item description, item difficulty, and misfit item s. Open in a separate window. For example, item difficulty of item 1 for Girl is 0. Analyses Classical test theory Classical test theory procedures were conducted to determine item difficulty mean and item discrimination corrected item-total correlations.

Item response modeling Exploratory factor analysis was used to assess whether the data satisfied the unidimensionality assumption of item response modeling, and was conducted for each subscale using SPSS IBM Corp, Differential item functioning Differential item functioning is tested by a significant interaction term e.

Primary Sidebar

Results Descriptive statistics The online questionnaire was completed by parents, whose demographic characteristics are shown in Table 1. Table 1 Demographic characteristics of participants. Item response modeling model fit A chi-square test revealed a significant difference effective vegetable parenting practices: Item fit A summary of item difficulties and misfit indicators is shown in Table 2. Item-person fit Wright map The Wright maps are presented in Fig. Table 3 Item description and estimates of differential item functioning where significant. Socioeconomic status Parents from high income families tended to encourage their child to try a couple of bites of a vegetable, and to reward their child with sweets if they eat their vegetables.

Discussion The present study investigated both item difficulty based on multidimensional polytomous item response modeling using the partial credit model which identified misfit items, and examined items exhibiting differential item functioning on the basis of demographic subgroups for a vegetable parenting practices instrument. Conclusion Limitations of an existing vegetable parenting practices scales were identified. Footnotes Conflict of interest: A didactic explanation of item bias, item impact, and item validity from a multidimensional perspective.

Journal of Educational Measurement. Revising translated differential item functioning items as a tool for improving cross-lingual assessment. Applied Measurement in Education. Cross-cultural equivalence of feeding beliefs and practices. The psychometric properties of the child feeding questionnaire among Blacks and Hispanics.

Constructing Measures : An Item Response Modeling Approach

A rating formulation for ordered response categories. Application of a psychometric rating model to ordered categories which are scored with successive integers. Perspectives on differential item functioning methodology.

Holland PW, Wainer H, editors. Does participation in an intervention affect responses on self-report questionnaires? Dimensions of vegetable parenting practices among preschoolers. Patterns of parental authority and adolescent autonomy. New Directions in Child and Adolescent Development. Significance tests and goodness of fit in the analysis of covariance structures. Differential item functioning DIF analysis of computation, word problem and geometry questions across gender and SES groups.

Studies in Educational Evaluation. Issues in studying close relationships. Conceptualizing and measuring closeness. Child feeding practices and the etiology of obesity. Obesity Silver Spring, Md. Confirmatory factor analysis of the Child Feeding Questionnaire. A measure of parental attitudes, beliefs and practices about child feeding and obesity proneness. Estimating item parameters and latent ability when responses are scored in two or more nominal categories.

Vegetables and fruit in the prevention of chronic diseases. European Journal of Nutrition. Applying the Rasch model. The home environments of children in the United States part I. Measuring the Mind Denny Borsboom. Rasch Models for Measurement David Andrich. Introduction to Psychometric Theory Tenko Raykov. Handbook of Test Development Mark R. Latent Variable Models John C. Nonparametric Statistics Dale I.

Bestsellers in Research Methods: Decolonizing Methodologies Linda Tuhiwai Smith. Doing Visual Analysis Per Ledin. Doing a Systematic Review Angela Boland. Reflexive Methodology Mats Alvesson. A Practical Guide Helen Aveyard. Researching Psychotherapy and Counselling Rudi Dallos.

Timelines of History Smithsonian Institution. Successful Qualitative Research Dr. The Basics Angela Goddard. Naked Statistics Charles Wheelan. Action Research Jean McNiff. Research for Designers Gjoko Muratovski. Visual Methodologies Gillian Rose. Write to Market Professor Chris Fox. Social Research Methods Alan Bryman.

Nursing Research Helen Aveyard. Creative research methods in the social sciences Helen Kara. How to Research Malcolm Tight. Analysing Discourse Norman Fairclough. Table of contents Contents: A Constructive Approach to Measurement. The "4 Building Blocks" Approach. The Four Building Blocks. Choosing and Evaluating a Measurement Model.

• Please Dont Dance In My Ambulance
• My Reminiscences of the Civil War
• Tipps und Tricks für den Anästhesisten: Problemlösungen von A bis Z für die Anästhesie, Intensivmedi
• Its Just a Matter of Time
• Book of Dragons: Volume Three