Most coaching literature has stressed where weight should be positioned at different events in the swing (e.g. Leadbetter, 1993). However, only two of the five scientific studies that have reported examining these factors found either differences in weight position between skill levels or that weight position was related to a performance measure. Further, no two similar studies have reported significant results for position of weight at the same swing event. For example, while Robinson (1994) found a relationship between weight position at TA and club velocity at ball contact, Mason et al. (1995) found no significant effect for a similar analysis at p < 0.05. In comparing handicap groups, Wallace et al. (1990) found one high and one low handicap golfer
differed at TA, MB, MD, BC and EF. However, Richards et al. (1985) found no statistical difference between a high and a low handicap group (N = 10, p < 0.05) at TB or BC, although medium effect sizes existed (calculated by this researcher and discussed below).
The only event that has been indicated as important in examining weight transfer and performance (TA, Robinson, 1994) has not been supported in other studies, providing conflicting and as a result inconclusive data. Robinson, using 30 golfers from a wide range of skills (professional to high handicap), found an association between the vertical force on the back foot at takeaway and club velocity at ball contact (r = 0.45, p = 0.02). This indicated that a larger Fz on the back foot, or positioning weight further towards the back foot, was associated with a larger club velocity. Robinson also reported that professional players tended to adopt a more balanced position of 51% on the back foot, while amateur players adopted a stance with weight further towards the front foot (only 42% on the back foot). No statistical analysis was reported for this comparison. These values were reported relative to the front foot in table 2.2 to make them consistent with other studies but have been reported here as they were presented by Robinson. Mason et al. (1995) found no association between club velocity at ball contact and the position of CP between the feet at TA for 64 apprentice professional golfers. It is possible that the small range of skill levels in the golfers tested influenced this result as a more homogenous sample may lead to a smaller effect size (e.g. Coleman, 1999). No other study has reported the relationship between the position of weight at TA and performance.
The comparison of weight position at TA between skill levels has been limited and might also be conflicting. Wallace et al. (1990) reported finding significant
differences at TA between skill levels using one low and one high handicap golfer. Wallace et al. reported that the low handicap golfer positioned CP closer to the front foot at TA (low = 63%; high = 49%; one way ANOVA, N = 10 shots, p < 0.001: no other statistics such as effect size or F-ratios were reported for a more precise
evaluation). It should be noted that both golfers used in the Wallace et al. study would have been part of the amateur group in the Robinson (1994) study, so for this reason alone the group comparison aspect of these two studies cannot be compared. No other study has reported differences between skill levels for TA.
Only one study has reported finding statistical support for the importance of weight position at swing events other than at TA. Wallace et al. (1990) examined one high handicap and one low handicap golfer performing 10 swings with the driver. CP between the feet was quantified at six swing events: TA, MB, TB, MD, BC and EF. CP at all swing events except TB was significantly different between the golfers. However, these findings have not been supported by other research. Robinson (1994) found no significant relationships between club velocity and Fz% at TB, FH, BC and MF (MF was defined as 0.25 s after BC). Robinson expressed surprise in finding no association between club velocity and the position of weight at ball contact, noting that moving the weight to the front foot at ball contact is a swing characteristic emphasised in golf instruction and swing measurement devices. Similarly, Mason et al. (1995) found no significant associations between club velocity and CP between the feet at TA, TB, BC and EF. As well, Richards et al. (1985) found that the position of centre of vertical forces (COV) at TB and BC did not differ between high (> 20) and low (< 10) handicap players.
It is important to note that low statistical power and limited generalisability due to small subject numbers existed in all weight transfer studies. Wallace et al. (1990) used only N = 2 subjects making it inappropriate to generalise the study’s findings. Also affected by low N, the Robinson (1994; N = 30) study possessed low statistical power. To achieve the 80% power recommended by Cohen (1968; i.e. 80% chance that a significant relationship will be found if it exists), Robinson (1994) would have needed to find large effect sizes among the relationships tested (r > 0.5; for the study’s parameters of p = 0.05 and N = 30). As the study was presented, the minimum detectable effect was r = 0.35, and so small effects were not detectable (using levels defined by Cohen, 1968; small effect: r > 0.2, moderate effect: r > 0.3, large effect: r > 0.5).
Richards et al. (1985) offset the loss in statistical power due to low N by increasing the expected effect size. Richards et al. compared groups that were further apart in skill level than all other studies (Low Handicap 0-10; High Handicap 20+). However, the effect size was decreased due to a large variance in the high handicap group and so the attempt to increase power by choosing skill levels that differed more than in other studies was offset by the less skilled group being highly variable in
performance. For the Richards et al. study, effect sizes were medium (calculated by this researcher using Cohen 1988, small: d = 0.2, medium: d = 0.5, large: d = 0.8), for the difference between groups at TB (effect size d = 0.51) and BC (effect size d = 0.73). With a medium effect size and alpha set at p = 0.05, more than 25 subjects in each group were needed to achieve 80% power. With only N = 10 as used in Richards et al., even a large effect size (effect size d > 0.8) would have possessed only 40% power (i.e. there is less than an even chance of finding a significant result if it exists).
Mason et al. (1995) collected a larger sample (N = 64) but the narrow range of skill levels reduced statistical power (PGA apprentice professionals only). Small effect sizes can be expected in elite level groups as the difference between subjects will be small (e.g. Ball et al., 2003a; Ball et al., 2003b; Coleman, 1999), and this will in turn reduce statistical power (e.g. Aron and Aron, 1999). The power to detect small effects in the Mason et al. (1995) study was only 36%. Further, even with N = 64, some small effects would not have been detected in the study as, at p < 0.05, an r-value of 0.24 was required for significance.