Potential barriers to walking were identified through open-ended interviews and focus groups with adolescents and young adults (ages 15–22 years). To elicit thoughts about barriers to walking, focus group participants were presented with descriptions of situations that they were likely to encounter in their everyday lives. Participants were then asked about factors that may prevent walking in those situations. Frequently-mentioned barriers were converted into scale items and pilot tested among young adults for clarity and coherence. After minor revisions and rewording, twelve items were developed for a questionnaire. These items assess the extent to which barriers — including concern over perspiring in nice clothing, fear of ruining hairstyle, fear of ruining nice clothing, wearing restrictive clothing, foot pain, blisters, uncomfortable shoes, having a lot to carry, time constraints, a lack of sidewalks, and hot and cold weather — impede walking for physical activity. To clarify the targeted level of physical activity, all items are prefaced by the following question: "Thinking about the past 3 months, how much do the following things prevent you from accumulating at least 30 minutes of walking throughout your daily routine?" Responses to items were provided using a 4-point Likert-type scale, with 1 = not at all, 2 = slightly, 3 = somewhat, and 4 = a great deal. A response of does not apply was also available.

The sample was composed of 305 undergraduate students from a major university in southern California. Participants ranged in age from 18 to 46 years with a mean age (± SD) of 20.6 (± 3.0), and 70.3% were female. The sample reported their overall health to be good, with a mean (± SD) response of 3.5 (± 0.9) on a 5-point scale; average body mass index (BMI) in kg/m² was 22.6 (± 4.0). BMI was calculated from self-reported height and weight. The ethnic breakdown of the sample was 49.8% Asian-American or Pacific Islander, 18.8% white, 15.8% Hispanic, and 15.6% other.

Perceived barriers to regular vigorous-intensity physical activity were measured through an instrument developed by Allison et al (18). The instrument is a scale composed of 16 items; nine items reflect internal barriers such as lack of energy, lack of self-discipline, and feeling stressed, and seven items reflect external barriers such as time constraints and cost. The perceived barriers instrument uses a 5-point response scale from 1 (not at all) to 5 (a great deal). In this study, internal consistency for the internal barriers subscales (Cronbach α = .82) and external barriers subscales (Cronbach α = .76) was satisfactory. Although this instrument was originally designed to assess perceived barriers to vigorous exercise among high-school students, it was determined to be the most relevant instrument of its kind for college-aged students. 

A dual-mode accelerometer (Manufacturing Technology, Inc [MTI], Ft Walton Beach, Fla, model 7164) provided an objective measurement of physical activity. The device is small (5.1 x 4.1 x 1.5 cm, 14g) and collects activity data unobtrusively because it is attached to a nylon belt worn around the waist. An analog bandpass filters the instrument's acceleration signal, which is subsequently digitized by an eight-bit analog-to-digital converter at a sampling rate of 10 samples per second and stored in 1-minute intervals (20). Once these data were downloaded, a data summary program (Actisoft Analysis Software, MTI, Ft Walton Beach, Fla) calculated time (in minutes) spent in moderate activity (e.g., brisk walking, 3–6 metabolic equivalent values [METs]) using the original cut-off points (1951–5724 counts per minute) established by Freedson et al (21). To adjust for differences in total monitoring time, the percentage of total minutes spent in moderate activity was computed. Accelerometers were worn for 4 consecutive days. This monitoring period has been determined to produce results that are at least 80% reliable (22). Participants with fewer than 48 hours of total monitoring for the 4 days (i.e., 12 waking hours per day) were excluded from the analyses.

Walking for transportation and walking for recreation were measured using a 3-Day Physical Activity Recall (3DPAR) validated by Motl et al (23). Participants recalled their activity for the previous 3 days between 7:00 am and 11:30 pm, segmented into 30-minute intervals. The number of 30-minute intervals of walking for transportation and walking for recreation were counted across the 3 days. Because of the substantial number of participants who reported no walking for transportation and no walking for recreation during the 3-day period, these variables were both dichotomized into some or no walking.

Participation in lifestyle physical activities was assessed with the Stanford Usual Physical Activity Questionnaire (24). Participants reported their usual participation in six lifestyle activities using a yes–no scale. These activities included taking the stairs instead of the elevator, walking short distances instead of driving, parking away from a destination in order to walk more, walking during lunch or after dinner, getting off at a bus stop before a destination and walking, and other extra walking or stair climbing for exercise.

College students volunteered for the study in exchange for course credit. A sample of 305 students completed the walking and vigorous exercise barriers instruments, 3DPAR, and Stanford Usual Physical Activity Questionnaire. Of these individuals, a random subsample of 104 participants was asked to complete the walking barriers instrument a second time after 5 days and to wear the MTI accelerometer device for the 4 intervening days. Ninety-six participants agreed to wear the accelerometer, of which 88% (n = 85) met the minimum compliance criteria. Because of the limited number of MTI monitors available, it was not feasible to obtain accelerometer data from all of the original study participants. However, the subsample did not significantly differ from the larger sample on any of the key variables of interest (e.g., perceived barriers and physical activity levels [data not shown]). The University of California, Irvine Institutional Review Board approved the procedures, and all participants provided written informed consent. Data were collected during 2003.

Data were screened for violations of statistical assumptions (e.g., normality, linearity) before the analyses. The percentage of MTI monitoring time spent in moderate-intensity physical activity was positively skewed and thus was subjected to a square root transformation. Missing data were handled with listwise deletion for principal components analysis (PCA) and hierarchical regression and pairwise deletion for all other statistical analyses.

PCA was performed to investigate the underlying factor structure of the perceived barriers to walking items. PCA is a statistical technique used to uncover coherent and independent groups of variables within a larger set (25). These groups of variables are called factors. In PCA, variables within each factor are more highly correlated with each other than with other factors. For this study, PCA was used to understand which walking barriers should be grouped together. To determine how the variables should be combined, PCA assigns values to each variable that reflect the extent to which it is related to each factor (i.e., factor loadings). The current study used an orthogonal rotation of the factor axes (varimax) to maximize the variance of the squared loadings of a factor on all the variables. This approach tends to maximize higher factor loadings and minimize lower factor loadings to yield results that facilitate the identification of each variable with a single factor.

Two separate PCAs were conducted. The first was used to determine the factor structure of the barriers to walking instrument and to identify any items that did not contribute to the integrity of the instrument. The second, in which the items from the walking barriers instrument were subjected to PCA together with the items from the vigorous activity barriers instrument, functioned to confirm that these two instruments were, in fact, measuring distinct constructs.

Further analyses evaluated the reliability and validity of the walking barriers items. The test–retest reliabilities of the walking barriers factors were evaluated through intraclass correlations of scores from each test administration. Pearson bivariate correlations assessed the factors for discriminant validity (i.e., divergence from the vigorous exercise barriers scales). Criterion validity (i.e., convergence with physical activity indicators) was evaluated with t tests comparing groups of individuals indicating some vs no walking for transportation and recreation on the 3DPAR. T tests also compared participants reporting yes vs no to the questions on the Stanford Usual Physical Activity Questionnaire. Criterion validity was further assessed by using objective measurements of moderate-intensity physical activity (via the MTI accelerometer). A stepwise hierarchical regression was used to determine if the walking barriers factors (entered in the second step) explained unique variation in moderate-intensity physical activity (measured by MTI accelerometer) after controlling for internal and external barriers to vigorous exercise (entered in the first step).

On average, participants wore the accelerometer about 14.5 hours on each of the 4 days of monitoring (data not shown). During this time, they engaged in an average (± SD) of approximately 29.9 (± 17.1) minutes of moderate-intensity physical activity per day. On the 3DPAR, participants reported an average of 18.2 (± 31.0) minutes of walking for transportation and 8.2 (± 20.0) minutes of walking for recreation per day. In total, 47.4% of participants reported some walking for transportation, and 25.0% of participants reported some walking for recreation during the past 3 days. Among participants, 80.6% usually took the stairs instead of the elevator, 67.8% usually walked short distances instead of driving, 25.0% usually parked away from a destination in order to walk more, 24.3% usually walked during lunch or after dinner, 4.6% usually got off at a bus stop before their destination in order to walk more, and 52.0% usually performed extra walking or stair climbing for exercise.

Means and SDs for the walking barriers items are shown in Table 1. In general, the mean ratings for the barriers items were low to moderate (i.e., scores were between 1 and 2 on a 4-point response scale). Lack of time, having a lot to carry, and wearing uncomfortable shoes were rated the highest. Blisters, concern over ruining one's hairstyle, foot pain, and lack of sidewalks were considered to pose the least hindrance to walking for physical activity.

In the PCA of the barriers instrument, the first iteration of the PCA found that two of the barriers to walking, hot weather and cold weather, did not significantly load onto any of the factors. Thus, these two items were removed, and the PCA was rerun. The second iteration of the PCA found that all of the walking barriers items significantly loaded onto one of three factors, which accounted for 61.96% of the variance. The rotated factor loadings of the items are shown in Table 1. Loadings greater than .50 indicated significant factor identification, as this suggests at least a 25% variance overlap between the item and the factor (26). The four items clearly loading onto the first factor were barriers related to personal appearance (i.e., perspiring, ruining nice clothing, ruining hairstyle, restrictive clothing). Three perceived barriers pertaining to footwear (i.e., uncomfortable shoes, blisters, foot pain) loaded onto the second factor. The third factor consisted of three situational barriers (i.e., lack of time, having a lot to carry, and lack of sidewalks).

In the PCA of the walking barriers and vigorous activity barriers instruments combined, the number of factors was constrained to two in order to reflect the major underlying conceptual differences between walking and vigorous exercise barriers. Results found that the 16 vigorous exercise barriers loaded significantly onto the first factor, and the 10 walking barriers items loaded onto the second factor. The majority of the factor loadings were greater than .50. There was no significant overlap in item loading between the two factors (data not shown).

Internal consistency and test–retest reliabilities for the walking barriers factors were acceptable. Chronbach α values were as follows: 0.82 for appearance barriers, 0.65 for footwear barriers, and 0.53 for situational barriers. Five-day test–retest reliability for the mean of the appearance barriers was 0.91 (95% CI, 0.87–0.94); for the mean of the footwear barriers, 0.71 (95% CI, 0.55–0.81); and for the mean of the situational barriers, 0.77 (95% CI, 0.65–0.85).

Discriminant validity of the walking barriers factors was assessed through relationships with the vigorous physical activity barriers scales. Table 2 shows the intercorrelations between walking barriers and vigorous exercise barriers. The correlations between factors within each construct (e.g., appearance and footwear barriers to walking, r = .497; internal and external barriers to vigorous exercise, r = .614) were generally larger than the correlations between factors from different constructs (e.g., situational barriers to walking and internal barriers to vigorous exercise, r = .179; footwear barriers to walking and external barriers to vigorous exercise, r = .230). 

To evaluate the criterion validity of the walking barriers factors, we first examined relationships with objectively measured physical activity. Bivariate correlations showed that appearance and situational walking barriers were significantly negatively related to moderate-intensity activity as measured by MTI accelerometer (Table 2).

The association of perceived barriers to participation in lifestyle activities and walking for transportation and recreation was also examined (Table 3). Based on the Stanford Lifestyle Activity Questionnaire, appearance barriers to walking were significantly associated with taking the stairs instead of the elevator and walking instead of driving. Situational barriers to walking were related to taking the stairs instead of the elevator and walking at lunch. Similarly, situational barriers were related to walking for transportation as assessed by the 3DPAR. In contrast, walking for recreation, assessed with the 3DPAR, was associated with neither appearance barriers nor situational barriers to walking.

A hierarchical regression was conducted in order to determine if walking barriers predicted moderate activity above and beyond the effects of vigorous exercise barriers. Footwear barriers to walking were not included in the regression model because they were unrelated to physical activity criteria in bivariate analyses (Table 2). The step including the appearance and situational barriers to walking explained a significant proportion of variation in moderate-intensity physical activity after controlling for internal and external barriers to vigorous activity (change in R ² = .121 for the second step, P = .01). Internal and external vigorous exercise barriers were unrelated to moderate-intensity physical activity after taking into account the effects of the walking barriers. The entire model explained 13.9% of the variance in the percentage of time spent in moderate-intensity physical activity.