Problem solving is described as “the self-directed cognitive-behavioral process by which a person attempts to identify or discover effective or adaptive solutions to problems encountered in everyday living” (D’ Zurilla and Nezu, 1999). Numerous studies have shown that problem-solving ability is associated with adjustment outcomes (D’ Zurilla et al., 1998; Elliott, 1999; Dreer et al., 2005a,b; Jaffee and D’ Zurilla, 2009). For example, good problem-solving skills (e.g., rational problem solving and positive problem orientation) are associated with fewer physical symptoms (Elliott and Marmarosh, 1994), higher life satisfaction (Dreer et al., 2005a), and better adjustment and social competence (Heppner and Anderson, 1985; Nezu, 1985; Cheng, 2001). Less constructive problem solving is associated with depression, anxiety, and emotional distress (Miner and Dowd, 1996; Cheng, 2001; Kurylo et al., 2004; Dreer et al., 2005a). Similarly, interventions that successfully improve problem-solving skills have been shown to reduce symptoms of anxiety, depression, and posttraumatic stress in people with chronic mental or medical health conditions; (Nezu et al., 1998, 2003; Perri et al., 2001; Ciechanowski et al., 2004; Klein et al., 2011; Ghahramanlou-Holloway et al., 2012) and in their caregivers (Bucher et al., 1999; Sahler et al., 2002, 2005, 2013; Cameron et al., 2004).

The comprehensive 52-item Social Problem-Solving Inventory-Revised (SPSI-R) (D’ Zurilla et al., 2002) is a widely used measure for interventions that target problem-solving ability (Varni et al., 1999; Cameron et al., 2004; Ciechanowski et al., 2004; Askins et al., 2009; Iobst et al., 2009; Klein et al., 2011; Sahler et al., 2013). The SPSI-R consists of five scales: Positive Problem Orientation (PPO); Negative Problem Orientation (NPO); Rational Problem Solving (RPS); Impulsive/Carelessness Style (ICS); and Avoidance Style (AS). The RPS scale can be divided further into four subscales: Problem Definition and Formulation; Generation of Alternative Solutions; Decision Making; and Solution Implementation and Verification. A shorter 25-item version of the SPSI-R measures the same five problem-solving scales but does not divide the Rational Problem-Solving Scale into subscale scores (D’ Zurilla et al., 2002). Finally, a 10-item version of the SPSI-R, consisting of one total problem-solving scale score, was developed by Dreer et al. (2009).

The 52-item SPSI-R domains exhibit adequate to excellent internal consistency for numerous populations and settings (D’ Zurilla et al., 2002; Askins et al., 2009; Iobst et al., 2009; Jaffee and D’ Zurilla, 2009; Klein et al., 2011; Wang et al., 2013). The 52-item version has been translated and validated for the Spanish-speaking population (Maydeu-Olivares et al., 2000) and has been culturally validated for the Hispanic population in the United States (De La Torre et al., 2010). Although it has been used to a lesser extent, the 25-item SPSI-R maintains satisfactory internal consistency in several populations (D’ Zurilla et al., 2002; Cameron et al., 2004; Hawkins et al., 2009; Li et al., 2016). The factor structure has been studied for both the 52-item (Maydeu-Olivares and D’ Zurilla, 1996; D’ Zurilla et al., 2002; Wang et al., 2013) and the 25-item (D’ Zurilla et al., 2002; Hawkins et al., 2009; Li et al., 2016) versions. Most studies confirmed the five-factor structure (Maydeu-Olivares and D’ Zurilla, 1996; D’ Zurilla et al., 2002; Hawkins et al., 2009), with only a few exceptions (Wang et al., 2013; Li et al., 2016). Maydeu-Olivares et al. (2000) replicated the 5-factor structure for the Spanish 52-item SPSI-R and showed language factorial invariance (i.e., equal factor loadings between the two language groups) across all domains, except for the impulsivity/carelessness scale. The 10-item version of the SPSI-R was developed in a single study of three clinical samples of adults (Dreer et al., 2009). Dreer et al. (2009) reported an acceptable person separation reliability [i.e., a measure like Cronbach’s alpha (α)] of 0.72 and demonstrated equivalency between the 10-item and 25-item SPSI-R versions. However, no studies have simultaneously assessed the psychometric properties of the 52-, 25-, and 10-item SPSI-R. In addition, research on the sensitivity to detect change is limited. Sensitivity to change is a critical property to evaluate because the SPSI-R is often used in intervention research. A widely used method to assess a measure’s ability to detect change involves randomized trials in which interventions of known effectiveness are compared with placebo or alternative approaches (Stratford and Riddle, 2005).

As far as we know, no other studies have looked at the psychometric properties of the 25- or 10-item SPSI-R version across different languages (i.e., Spanish and English). The goal of the current study was to examine the factor structure and factorial invariance of the 52-, 25-, and 10-item versions of the SPSI-R across English- and Spanish-speaking caregivers of children with cancer. Next, we tested (a) internal consistency, (b) test-retest reliability, and (c) sensitivity to detect changes in problem-solving skills over time for the 52-, 25-, and 10-item versions of the SPSI-R using real-world data from three randomized clinical trials of a problem-solving skills training intervention (in these specific cases, Bright IDEAS; Sahler et al., 2005, 2013; Phipps et al., 2020). These data inform whether study participant burden could be decreased by using shorter versions of the SPSI-R in English- and Spanish-speaking caregivers of children with cancer without losing critical effect data.

Two separate CFA’s (one for the 52-item version and one for the 25-item version) were conducted in Mplus version 7.4 on the five a priori factors of problem-solving. Another CFA was conducted on the 10-item version, with one a priori defined total factor of problem-solving skills. The CFA’s for the 52- and 25-item versions showed adequate model fit, whereas the CFA for the 10-item version showed insufficient model fit (Table 2). Therefore, subsequent analyses were only performed with the 52- and 25-item version. Multigroup CFA’s revealed metric and scalar invariance (i.e., ΔCFI < 0.01) across time (T1 versus T2) and language (English versus Spanish) for the 52- and 25-item versions of the SPSI-R (Table 2). The 52- and 25-item versions of the SPSI-R showed sufficient to excellent internal consistency (α = 0.71–0.94; Table 3) for the total scale and all its subscales. The test-retest reliability of the 52- and 25-item SPSI-R versions was stable over time (r = 0.52–0.77; Table 4). Both versions of the SPSI-R detected significant changes in problem-solving skills over time between participants assigned to the problem-solving intervention (PSST) and control (UPC/NDS) groups (Table 4). Participants in the intervention group demonstrated significantly higher changes in their problem-solving skills from T1-T2 on all SPSI-R domains than did participants in the control group (p < 0.00–p < 0.05), except for the impulsivity/carelessness subscale (p = 0.153–0.132).

The current study assessed the psychometric properties of 3 published versions of the SPSI-R (52-, 25-, and 10- items) in English- and Spanish-speaking caregivers of children with newly diagnosed cancer. Our findings indicate that the 52- and 25-item versions showed acceptable internal consistency for both the English and Spanish versions. This result agrees with numerous other studies using the 52- and 25-item versions of the SPSI-R (Chang, 2002; D’ Zurilla et al., 2002; Cameron et al., 2004; Hawkins et al., 2009; Jaffee and D’ Zurilla, 2009; Klein et al., 2011; Pech and O’Kearney, 2013; Wang et al., 2013; Li et al., 2016). We were not able to reproduce the 1-factor structure of the 10-item version in the current population and were therefore not able to demonstrate acceptable reliability for the 10-item version. This result is in contrast to the one other study using the 10-item version on other populations (Dreer et al., 2009). Reverse worded items are often used to reduce or eliminate acquiescence bias. The 10-item version of the SPSI-R consisted of 7 reverse worded items and 3 original positively worded items. It is known from the literature that the number and type of reverse worded items can problematically affect factor structures of measurement instruments (e.g., Zhang et al., 2016). Future research should examine the origin of the items selected for the 10-item version of the SPSI-R. Multigroup analyses showed that the factor structure of the 52- and 25-item versions of the SPSI-R are invariant across time and language, showing equal factor loadings and intercepts at two time points and for English- and Spanish-speaking caregivers. Test-retest reliability showing large correlations for the comparison groups between T1 and T2 suggest that the 52- and 25-item versions of the SPSI-R were relatively stable over time. The psychometrics of the SPSI-R 52- and 25-item versions are similar for English- and Spanish-versions. This corresponds with one other study measuring the psychometrics of the 52-item version in Hispanics (De La Torre et al., 2010), and is partly in line with a previous study that tested for factorial invariance of the 52-item version in the Spanish population (Maydeu-Olivares et al., 2000). Whereas the current study demonstrated factorial invariance across all domains of the SPSI-R, Maydeu-Olivares et al. (2000) reported invariance for all domains except for the impulsivity/carelessness scale.

The 52-, and 25-item versions of the SPSI-R were sensitive to change over time, as they both detected the expected differences in problem-solving skills between participants assigned to the intervention versus the comparison groups, except for the impulsivity/carelessness subscale. One possibility for not finding pre-post changes in this subscale is floor effects. Floor and ceiling effects can influence a scale’s responsiveness to change because they limit our ability to measure variance above or below a certain limit (Cramer and Howitt, 2004). The population broadly scored low on this scale to start and therefore there was not much space to improve on this domain. Since this domain includes items that describe impulsively making choices without thinking things through, it’s possible that there is some social desirability bias to respond low on this domain. The treatment program focuses on improving rational problem-solving, so it does make sense that we would see the greatest improvement there. Moreover, the SPSI-R 52- and 25-item versions demonstrated the same degree of differences over time for all subscales. We conclude that the 25-item version is a psychometrically reasonable substitute for the 52-item version if RPS subscales are not required. If the RPS subscales are not needed, the 25-item version can serve as a rapid assessment of problem-solving skills with less burden for the participant.

Although our findings unequivocally demonstrate strong psychometric properties of the SPSI-R, some limitations of our study should be mentioned. First, because the first two trials of the problem-solving skills intervention included only mothers, this population was overrepresented. Future studies are needed to determine whether the psychometrics of the SPSI-R are robust for fathers. Because of the limited number of fathers in our sample, we were not able to test for factorial invariance across gender in this population. Second, our findings are limited to caregivers of children with cancer. Future work needs to expand our findings with additional groups to determine generalizability.

In conclusion, we have demonstrated that the 52- and 25-item versions of the SPSI-R are psychometrically sound measures to assess problem-solving ability in both English and Spanish caregivers of children with cancer. The 25-item version is a reasonable substitute for the 52-item if the subscales of the RPS are not required. The 10-item version of the SPSI-R is not a psychometrically sound substitute for the longer versions of the SPSI-R either for assessment of change or for screening problem-solving skills.

The raw data supporting the conclusions of this article is confidential data and can be made available upon reasonable request.

The studies involving humans were approved by the Internal Review Board of the University of Rochester School of Medicine and Dentistry. IRB protocol numbers: RSRB 05736 2002 trial; RSRB 09840 2005 trial; RSRB 38174 2013 trial. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

SS: conceptualization, methodology, writing original draft, formal analysis, and visualization. SP, RN, and OS: funding acquisition, conceptualization, methodology, data collection, supervision, and writing review and editing. KD: conceptualization, methodology, data collection, supervision, resources, and writing review and editing. DF: methodology, formal analysis, validation, and writing review and editing. MD, KI, NS, and MA: conceptualization, methodology, data collection, supervision, and writing review and editing. MV: data collection, project administration, and writing review and editing. All authors contributed to the article and approved the submitted version.