Accuracy and consistency of weights provided by home bathroom scales

Self-reported heights and weights are often used in public health research with adults, children and families because these data are easy and inexpensive to collect [1]. However, self-reported and expert-measured weights may differ by factors such as age, sex, and perceived weight status [2–4]. In addition to parents reporting their own weight, they also may be asked to report the weight of their children. A child’s weight status as well as parental perception of their child’s weight affect accuracy of reports [4, 5]. A recent study reported that parents of overweight children ages 2 to 6 years old erroneously overestimated their child’s weight, but parents of older overweight children and adolescents underestimated their child’s weight [4]. Similar findings were reported using data from two nationally representative surveys [6]. Others have noted mothers underestimate child weight [7, 8].

Inaccuracies in reported weights often are attributed to social desirability and/or erroneous measurements or recalls [1, 9, 10]. Minimal research, however, has examined the accuracy of a common tool used to measure self-reported weights, that is, home bathroom scales. The limited available data suggest that home scales, as well as medical grade scales used by physicians, can vary in accuracy and precision [11]. A study of 37 dial-type bathroom scales in British clinics reported inaccuracies of more than 1% compared with a calibrated electronic scale, suggesting that digital scales may be more accurate [12]. Further, an evaluation of 233 scales (type not specified) from United States primary care, diabetology and endocrinology clinics, and fitness and weight loss centers found that more than a quarter of the scales were more than 0.9 kg imprecise when tested with a 45.5 kg standard weight. At 113.6 kg, about one in five scales was imprecise by more than 2.7 kg, or about 1 Body Mass Index (BMI) unit [10]. Several factors, such as type of flooring, foot placement on the scale, and type of clothing or shoes worn during weighing, may influence accuracy of scales [13].

Widespread implementation of community-based obesity prevention programs targeting children and families is currently underway [14–16]. Some of these programs, and the research used to develop them, rely on self-reported weight or BMI, which is calculated using weight and reported as a primary outcome or measure of intervention effectiveness [14, 17, 18]. A major limitation to establishing intervention (in)effectiveness is inaccurate weight reports [1, 19, 20]. To advance the work of public health professionals in implementing effective programming aimed at ameliorating the obesity epidemic, it is important to increase the accuracy of self-reports. Thus, the goal of this study was to assess the accuracy of home bathroom scales to better understand how their use in the common process of self-weighing may affect accuracy of self-reported weight data [21].

Of the 67 bathroom scales that were received, 6 scales were eliminated due to damage or improper functioning (i.e. did not register a value when tested with the calibration weights) or because they were not a home scale. Of the final sample (n = 61), 18 were dial (30%) and 43 (70%) were digital scales from 16 various manufacturers. All were new, like new, or had light wear. The bathroom scales’ precision increments were 0.045 kg, 0.091 kg., 0.15 kg, and 0.45 kg (21%, 38%, 13%, and 28%, respectively). The maximum weight capacities ranged from 123 kg to 181 kg, with the most common capacity being 136 kg (35%). Dial scales were significantly older (p = 0.042; mean age 6.0 (6.9 SD) years old) than digital scales (mean age 3.6 (2.3 SD) years old). When the scales were first observed, all digital scales displayed 0, whereas dial scales displayed a mean absolute weight of 0.95 (1.9 SD) kg, with a range of -0.45 to 7.9 kg.

All participants reported using their bathroom scale on hard flooring. Scales were used daily (21%), weekly (46%), or monthly (27%), and 5% used their scale yearly or less than once a year. Participants reported infrequently calibrating their scale; only 28% calibrated the scale each time or most of the time before using it. Of these participants, 64% had dial-type scales.

Mean weight between Rounds 1 and 2 differed significantly only for dial scales tested with a 75 kg calibration weight (p = 0.028). For all other test loads, scales consistently registered the same weight, regardless of the scale type and load. Per 45.4 kg, the difference in absolute weight registered between Round 1 and 2 averaged approximately 0.11 kg (range from 0.0 kg to 0.6 kg), with the largest errors for the 10 kg calibration weight.

Table 1 compares accuracy of the mean weight registered by the home scales to the known calibration weight tested. For dial scales only, significant differences of weights between the test load and the weight displayed on the home scales occurred for all calibration weights and both weighed researchers (p < 0.05 for all). In contrast, significant differences for digital scale accuracy occurred only with the 75 kg calibration weight (p < 0.05). Dial scales were significantly more imprecise than digital scales at all test weights (Table 2). Figure 1 visually displays the absolute weight imprecision across each calibration weight test load. Scales became increasingly less precise as weight load increased; however, imprecision as a percent of total weight test load was inversely related to weight load. Accuracy did not vary by age of the scale. There was no significant difference between dial scales ≤3 years old (n = 10) and those ≥3 years old (n = 8), and no differences between digital scales ≤3 years old (n = 25) compared to those ≥3 years old n = 18). Thus, scale type, not age, is likely the source of inaccuracy and imprecision.

Multinomial regression was used to examine associations between absolute weight imprecision (i.e., <0.45 kg., 0.45 to <0.91 kg., 0.91 to <1.8 kg., 1.8 to < 2.7 kg, and ≥2.7 kg.) and type of scale. The majority of scales were precise within 0.9 kg of the actual weight of the load tested, but the extent of imprecision increased as the tested weight increased. For instance, at 50 kg, less than 2% of all tested scales were off by at least 1.8 kg; whereas the proportion rose to nearly 14% when the 110 kg calibration weight was tested. Further, the overall imprecision of dial scales was significantly greater than that of digital scales at all weights, with nearly 17% having a precision error of at least 2.7 kg or 1 BMI unit at a test load of 99.8 kg or greater (p < 0.05).

Findings from this study indicate that home bathroom scales are consistent in the weights measured. Dial scales were significantly more imprecise than digital scales at all calibration weight test loads measured with digital home scale weights differing significantly at the 75 kg test load. The imprecision at the 75 kg test load likely is due to human error in recording of data (e.g. incorrectly recording the weight as measured by the scale) during scale testing. The finding that scale precision was significantly higher at all test loads for digital versus dial-type scales confirms previous reports of significant, positive correlations between scale precision and accuracy [10]. Although the reasons for the differences between dial and digital scale precision are not completely clear, digital scales have fewer moving parts to get out of alignment or become damaged and have mechanisms to automatically set the starting weight to zero [9].

Although dial scales were significantly more imprecise than digital scales, absolute imprecision tended to be within 0.91 kg of calibration weights, a level far below the threshold that would cause weight to result in an error of one BMI unit (i.e., approximately 2.7 kg) [22]. It is also within the ~0.5 kg fluctuation in body weight considered as normal daily variation in healthy adults [22, 23]. For children, expected daily variation in weight is about 1.5 ± 0.5% of their body weight or about 375 to 750 g for children who weigh 25 to 50 kg [20]. For very young children, a difference of 0.91 kg could place them in a different BMI-for-age percentile, potentially resulting in misclassification of their BMI [22]. The present study suggests that dial scales in particular lack accuracy at weights below 10 kg; thus, use of a dial scale to weigh small children should be avoided. Given that weight distribution did not affect scale accuracy, it may be feasible for parents with young children who have dial scales to derive child weight by subtracting their own weight from their weight while holding the child. Future research is needed to explore the accuracy and feasibility of this technique.

This study suggests that inaccuracies in self-reported weight likely are due in large part to human bias and/or reporting or recall errors and not the home bathroom scales. Other studies have found that mothers with less education are more likely to provide inaccurate self-reported weight [5]. The majority of participants had a digital scale, suggesting that these are more likely to be used in self-reported weight. In addition, the finding that all scales were used on hard flooring, such as tile, wood, or cement, indicates that consumers understand the effect of floor coverings on scale accuracy, or scales are used in bathrooms, which coincidentally have hard flooring. The infrequent calibration history, however, suggests that consumers could benefit from reminders to set scales to zero before taking weights. Alternately, self-zeroing digital scales can help overcome this potential source of measurement error.

These findings suggest that errors made in self-reported weights are more likely due to human error or social desirability than scale inaccuracy. Importantly, this study suggests that home bathroom scales, especially digital scales, provide sufficiently accurate and consistent results for use in public health research. Providing participants with instructions for calibrating their home bathroom scales before use may further improve the accuracy of self-reported weight. In addition, researchers can query participants about scale characteristics such as type of flooring, room in which scale is housed, presence of moisture or steam in the environment that may cause rusting (i.e. in a bathroom), participant ability to calibrate the scale, age of scale and signs of wear to aid in interpretation of the data collected. Future research should also develop and validate instructions for measuring height at home to improve the overall accuracy of self-reported BMI. Finally, researchers should determine whether other factors, such as socioeconomic status or participant age, affect the type of scale owned, the amount of wear, and the frequency with which it is calibrated.