An Empirical Analysis of District Court Claim

Construction Decisions,

January to December 2009

In the past decade, a number of studies have scrutinized the Federal Circuit’s rate of reversal of

district court claim construction rulings.1 To date, however, there has been little empirical research

focusing on district court claim construction decisions themselves. Although district court statistics

represent only a “slice in time” before appeal, they are nevertheless important to litigants and trial

counsel, who must make various tactical decisions and cost-benefit analyses at the district court level

long before considerations of appellate reversal rates come into play.

¶2 This article explores claim construction outcomes at the district court level, including patentee

win rates on an element-by-element basis, win rates for broad constructions versus narrow

constructions, and win rates by jurisdiction and type of argument. To accomplish this goal, we

examined 211 district court Markman decisions from 2009, which collectively construed a total of

1858 disputed claim terms.2 We culled data from these decisions and performed various statistical

analyses to arrive at the outcomes reported herein.

The sample analyzed for this study consists of 211 district court claim construction decisions

issued between January and December 2009. An effort was made to collect every such decision

within this time frame, but it is likely that some were not collected because they were either not

included in a searchable database or were integrated into another decision (e.g., a summary judgment

decision) that escaped our search parameters. In any event, we believe our dataset includes the

majority of district court Markman decisions handed down in 2009.

¶27 Individual claim constructions (on a limitation-by-limitation basis) were excluded from the

dataset if we could not determine from the decision itself what the parties’ proposed constructions

were. This occurred in only a relatively few instances, i.e., where the district court pronounced its

claim construction for a particular limitation without explaining how it got there or what the parties’

respective arguments were.

¶28 Individual claim terms were excluded (on a limitation-by-limitation basis) if the parties agreed to

the proper construction. In other words, the sample we analyzed consisted only of disputed claim

constructions.

As shown in Figure 1, of the 1858 disputed terms in the sample we analyzed, the district court

sided with the patentee 36.8% of the time and with the accused infringer 15.1% of the time. In the

remaining 48.1% of the sample, the district court declined to adopt either party’s proposed

construction.

Another factor of interest is the ratio of the claim construction win rate for patentees versus that

for accused infringers. Nationwide, we calculated this ratio to be 2.4 in our sample (i.e., 36.8% claim

construction win rate for patentees versus 15.1% claim construction win rate for accused infringers).

In other words, nationwide, patentees won their proposed claim constructions 2.4 times more often

than accused infringers did. The table below shows the patentee/accused-infringer claim

construction win ratio for each jurisdiction in which at least forty claim terms were construed.

We next analyzed claim construction win rates based on whether the winning party proposed the

broader or narrower construction. For this analysis, we excluded any terms (approximately 47% of

cases) where neither party’s construction was clearly broader or narrower than the other’s.

¶43 Overall (excluding ambiguous cases), patentees proposed the broader construction about 90% of

the time, with accused infringers proposing the broader construction the remaining 10% of the time.

The overall win rate for the party proposing the broader construction was 47.8%. This is shown in

Figure 4 below.

Figure 5 also shows that in nearly all these districts, broader constructions won far more often

than narrower constructions. The one exception is the District of D.C., in which the narrow

construction won 43.3% of the time versus the broad construction win rate of 33.3%. It is also

noteworthy that in three districts where the broad construction win rate was lower than the

nationwide average, a high percentage of claim terms resulted in neither the broad nor the narrow

construction winning, i.e., neither construction won. These districts are the Central District of

California (62.5%), the Southern District of California (56.4%), and the Eastern District of Virginia

(48.5%).

¶46 We next analyzed whether it made any difference whether it was the patentee or the accused

infringer who offered the broader construction. To do this, we looked at overall claim construction

win rates when the patentee offered the broader construction versus when the accused infringer

offered the broader construction.

As shown in Figures 6 and 7 above, when patentees proposed the broader construction, they

prevailed 48.8% of the time, versus 13.9% for accused infringers. In contrast, when accused

infringers proposed the broader construction, they prevailed 38.5% of the time, compared to 30.8%

for patentees.

We next analyzed what effect, if any, the length of a proposed construction had on the likelihood

that a district court would adopt it. For this analysis, we used a ratio of the number of characters in

the proposed construction to the number of characters in the actual limitation itself. We called this

the “construction-to-limitation” or “C/L” ratio. For example, if a party proposed to construe

“connected” (9 characters) as “joined or fused together, e.g., welded, bolted, riveted, glued, clamped,

or held together by magnetism” (102 characters), then the C/L ratio for that proposed construction

would be 102/9 or 11.3.

¶49 Because many proposed constructions are, in fact, short legal arguments, e.g., “no construction

needed,” “indefinite,” “ordinary meaning,” we eliminated those types of arguments from the data set

for this particular analysis. The remaining dataset consisted of 950 “head-to-head” proposed claim

constructions, i.e., where both parties put forth a textual construction for a given limitation.

We looked at the overall win rate in our modified data set of 950 “head-to-head” constructions

for the shorter versus the longer proposed construction.

As shown in Figure 8, in head-to-head contests between two proposed constructions, one

shorter and the other longer, the claim-construction win rate for the more concise construction was

32.3%, versus 24.0% for the longer proposed construction.

¶53 We next attempted to determine if there was an identifiable trend in the claim-construction win

rates for patentees and accused infringers based on the length of their proposed constructions. For

this, we divided proposed constructions into sixteen groups depending on their C/L ratio. We then

calculated the claim-construction win rate for all constructions (combining patentee and accused

infringer constructions) in each of these sixteen groups.

SUMMARY AND CONCLUSION

¶64 In this study, we analyzed the results of 1858 disputed claim constructions in 211 cases in which

Markman decisions were handed down in 2009. The results indicate that patentees won their

proposed claim constructions 36.8% of the time, compared to 15.1% for accused infringers.

¶65 On average, broad constructions won approximately three times more often than narrow

constructions (47.8% versus 15.6%). There is a noticeable disparity, however, depending on which

party proposed the broad construction. When patentees proposed the broader construction, they

won 3.5 times more often than the accused infringer’s narrower construction (48.8% versus 13.9%).

In contrast, when accused infringers proposed the broader construction, they won only 1.3 times as

often as the patentee’s narrower construction (38.8% versus 30.6%). These results suggest there may

be a systemic bias in favor of patentees’ constructions.

¶66 On average, more concise constructions tended to win more often than less concise

constructions, although this relationship was more pronounced in the region of C/L ratios of 7 or

less, and was also more pronounced in patentees’ proposed constructions than in accused infringers’

constructions.

¶67 “No construction necessary” was a more successful argument for patentees than for accused

infringers. Our data indicate that when patentees argued “no construction necessary” or “ordinary

James R. Barney and Charles T. Collins-Chase: An Empirical Analysis of 22

District Court Claim Construction Decisions, January to December 2009

http://stlr.stanford.edu/pdf/barney-collins-chase-claim-construction.pdf